1. Field of the Invention
The present invention is directed to a method and apparatus for reducing the peak power probability of a band limited Gaussian signal by clipping the signal to constrain its spectrum within error-shaped bounds. More particularly, the invention is directed to reducing the peak to average power ratio in signals having a spread power distribution, including Code Division Multiple Access (CDMA) signals.
2. Description of Related Art
One of the more expensive components in a radio transmitter is a power amplifier. A power amplifier receives an input signal and in response generates a significantly more powerful output signal. The complex ratio between the power of the output signal and the power of the input signal is the amplifier""s gain. Except for a magnitude gain in power, the output signal is desirably an accurate reproduction of the input signal, so that an information component in the input signal is accurately amplified in the output signal.
In practice, power amplifiers are highly non-linear devices and accurate amplification is achieved only when the instantaneous power of the input signal lies within a narrow domain. Whenever input signal power increases beyond the linear domain, the power amplifier generates a distorted output signal.
Two serious consequences flow from distorting the output signal. First, a noise component is introduced into the output signal and tends to obscure the information component, since the output signal is no longer an accurate amplification of the input signal. Second, the distortion generates spurious radio emissions both inside and outside the frequency band allocated to the radio transmitter, these spurious radio emissions interfering with radio transmissions from other transmitters. It is therefore desirable that a power amplifier be well matched to its input signal, such that the input signal instantaneous power remains within the narrow linear domain of the power amplifier.
To avoid distortion, one solution is to overspecify the power amplifier. One could use a power amplifier having a very large linear domain, one that could easily contain the average power level of the input signal and could even contain much higher input power peaks when they infrequently occurred. However, as previously mentioned, power amplifiers are expensive and it is therefore wasteful to overspecify this component.
Another solution is to pre-process the input signal to ensure that its peak power is always constrained within limits dictated by a smaller, cheaper power amplifier. In other words, the ratio of the input signal peak power to average power could be constrained to a certain limit. Unfortunately, such pre-processing is highly dependent upon the nature of the input signal.
One class of input signals that is particularly challenging to pre-process is Gaussian signals, which are characteristic of spread-spectrum communication signals, including code division multiple access (CDMA) signals. These signals have a substantially uniform average power distribution across a predetermined frequency range. It is therefore challenging to remove any portion of the signal without introducing distortion.
A conventional pre-processing technique includes sampling the input signal and then hard clipping all peak samples above a pre-defined threshold. The resulting clipped signal thus consists of both a desired signal and a clipping error signal. Since the error signal will smear outband emissions, a digital filter must be connected after the hard clipper to minimize the outband emissions. However, because this filter processes both the desired signal and the error signal it must satisfy both inband and outband requirements. The inband requirement is dictated by need to only minimally distort the desired signal, while the outband required is determined by the need to minimize spurious emissions. More particularly, the filter""s inband frequency response is desirably as flat as possible. As a result of all these constraints, the filter is complicated to implement and has a high gate count.
To improve clipping performance according to the conventional technique, a higher sampling rate and consecutive multiple clipping are used. A higher sampling rate reduces the number of over-threshold peak samples that escape processing; however, not surprisingly, filter complexity increases with the sampling rate. Clipping a signal multiple times as it propagates through a radio transmitter resists post-clipping peak regrowth but also increases circuit complexity several fold.
What is needed therefore is a relatively simple yet effective method and apparatus for constraining the power peaks of a Gaussian signal.
The present invention is directed to such a solution, including a solution where a band limited Gaussian signal is received at a power amplifier for transmission. In particular, the Gaussian signal might be a composite of a set of code division multiple access (CDMA) signals in a multi-carrier mode or a wideband direct sequence CDMA signal.
At either the baseband or the intermediate frequency stage of the transmitter, embodiments of the invention reduce the probability of the peak to average power ratio of the signal. In contrast to the conventional technique, which includes hard clipping an input signal and then filtering the resulting clipped signal, the clipped signal including both the input signal and an error signal, embodiments of the present invention filter only the error signal. First the error signal is shaped with a shaping filter to reduce close-in outband emission, and then the shaped error signal is subtracted from the spread spectrum signal. After this error shaped clipping, a loose postprocessing filter, either lowpass or bandpass, is used to further reduce the far end spurious emission level. As a result, the probability of peak to average ratio is reduced and the outband spurious emission level is reduced to a required level due to the shaping filter and postprocessing filter. Since the shaping filter only applies to the error signal, the inband does not need to be flat. Therefore, the shaping filter is much simpler to implement than a filter according to the conventional technique.
Therefore, according to one aspect of the invention, there is provided a method of reducing the peak power probability of a spread spectrum signal, including clipping the signal to constrain its spectrum within error-shaped bounds. The method desirably includes generating a clipping threshold signal, generating a clipping error signal responsive to both the clipping threshold and the spread spectrum signal, filtering the clipping error signal to produce a shaped error signal; and subtracting the shaped error signal from the spread spectrum signal. This technique makes multiple clipping practical, which helps further reduce the probability of the peak occurrence.
Preferably, the method includes delaying the spread spectrum signal to align its phase with the shaped error signal for subtraction.
The step of receiving a spread spectrum signal might include receiving a baseband signal. In such case, it is desirable that filtering the clipping error signal includes lowpass filtering.
Generating the clipping error signal might include receiving a second instance of the baseband signal, receiving a third instance of the baseband signal, scaling the third instance of the baseband signal, and subtracting the scaled third instance of the baseband signal from the second instance of the baseband signal.
Preferably, scaling the third instance of the baseband signal includes receiving a fourth instance of the baseband signal, determining an RMS value of the fourth instance of the baseband signal, determining a peak value of the fourth instance of the baseband signal, dividing the RMS value by the peak value to produce a scaling factor, and multiplying the third instance of the baseband signal by the scaling factor.
In contrast, where receiving a spread spectrum signal includes receiving an intermediate frequency signal, it is desirable that filtering the clipping error signal includes bandpass filtering.
Generating the clipping error signal might thus include: receiving a second instance of the intermediate frequency signal, receiving a third instance of the intermediate frequency signal, amplitude clipping the third instance of the intermediate frequency signal, and subtracting the clipped third instance of the intermediate frequency signal from the second instance of the intermediate frequency signal.
Amplitude clipping the third instance of the intermediate frequency signal might include: receiving a fourth instance of the intermediate frequency signal, determining a current RMS value of the fourth instance of the intermediate frequency signal, multiplying the RMS value by a scaling factor to produce a scaled RMS value, setting a clipping threshold proportionate to the scaled RMS value, and clipping the third instance of the intermediate frequency signal when it exceeds the clipping threshold.
According to another aspect of the invention, there is provided an apparatus for reducing the peak power probability of a spread spectrum signal having an error-shaped clipper for constraining the spectrum within error-shaped bounds.
The error-shaped clipper might further include: a clipping threshold signal generator for generating a clipping threshold signal, a clipping error signal generator connected to generate a clipping error signal responsive to both the clipping threshold signal received from the clipping threshold signal generator and the spread spectrum signal, a filter connected to the clipping error signal generator to receive the clipping error signal and to produce in response a shaped error signal, and a first summing junction connected to subtract the shaped error signal from the spread spectrum signal.
The apparatus might further include a time-delay loop for delaying the spread spectrum signal to align its phase with the shaped error signal for subtraction.
Where the spread spectrum signal is a baseband signal, it is desirable that the filter for filtering the clipping error signal includes a lowpass filter.
Preferably, the clipping error signal generator includes: a first coupler for coupling a second instance of the baseband signal, a second coupler for coupling a third instance of the baseband signal, a divider connected to the second coupler to divide the third instance of the baseband signal, and a second summing junction connected to the first coupler and the divider to subtract the divided third instance of the baseband signal from the second instance of the baseband signal to produce the clipping error signal.
In this context, the word xe2x80x9ccouplerxe2x80x9d does not denote a radio frequency coupler. In this context, the word xe2x80x9ccouplerxe2x80x9d has its more general connotation as a device for transferring a signal, including a digital signal, from one portion of a circuit to another.
Desirably, the divider includes: a third coupler for coupling a fourth instance of the baseband signal; an RMS detector connected to the third coupler to generate an RMS signal responsive to the RMS value of the fourth instance of the baseband signal, a peak detector connected to the third coupler to generate a peak signal responsive to the peak value of the fourth instance of the baseband signal, a first multiplier connected to the RMS detector and the peak detector to divide the RMS signal by the peak signal to produce a scaling factor signal, and a second multiplier connected to the first multiplier and the second coupler to multiply the third instance of the baseband signal by the scaling factor signal.
In contrast, where the spread spectrum signal is an intermediate frequency signal, it is desirable that the filter for filtering the clipping error signal includes a bandpass filter.
Preferably, the clipping error signal generator includes: a first coupler for coupling a second instance of the intermediate frequency signal, a second coupler for coupling a third instance of the intermediate frequency signal, a clipper connected to the second coupler to amplitude clip the third instance of the intermediate frequency signal to produce a clipped third instance of the intermediate frequency signal, a second summing junction connected to the clipper and the first coupler to subtracting the clipped third instance of the intermediate frequency signal from the second instance of the intermediate frequency signal to produce the clipping error signal.
It is preferable that the clipper include: a third coupler for coupling a fourth instance of the intermediate frequency signal, an RMS detector connected to the third coupler to generate an RMS signal responsive to the RMS value of the fourth instance of the intermediate frequency signal, a terminal for receiving a user-adjustable scaling factor signal, a multiplier connected to the RMS detector and the terminal to multiply the RMS signal by the scaling factor signal to produce a scaled RMS signal, a variable amplitude hard-clipper connected to the multiplier and the second coupler to clip the amplitude of the third instance of the intermediate frequency signal back to a threshold level corresponding to the scaled RMS signal.
According to yet another aspect of the invention, there is provided a code division multiple access (CDMA) transmitter apparatus, having: a plurality of CDMA modems, each of the plurality of CDMA modems having a respective input and output, a digital pre-processing stage having a plurality of inputs, an output, at least one error-shaped clipper having an input and an output, at least one digital intermediate frequency (IF) processor having an input and an output, and a summing junction having a plurality of inputs and an output, each of the plurality of digital pre-processing stage inputs being connected to a respective CDMA modem output, and the plurality of digital pre-processing stage inputs being connected to the digital pre-processing stage output through the at least one error-shaped clipper, the at least one digital IF processor, and the summing junction, a digital to analog converter (DAC) having an input and an output, the DAC input being connected to the digital pre-processing stage output, an analog intermediate frequency and radio frequency (IF and RF) processor having an input and an output, the IF and RF processor input being connected to the DAC output, and a power amplifier having an input and an output, the power amplifier input being connected to the IF and RF processor output.
Desirably, the at least one error-shaped clipper includes a plurality of error-shaped clippers, each of the plurality of error-shaped clippers having an input connected to the output of a respective CDMA modem, the at least one digital IF processor includes a plurality of digital IF processors, each of the plurality of digital IF processors having an input connected to the output of a respective error-shaped clipper, and each of the plurality of inputs of the summing junction is connected to the output of a respective one of the plurality of digital IF processors.
Alternatively, each of the plurality of inputs of the summing junction is connected to the output of a respective one of the plurality CDMA modems, the input of the at least one error-shaped clipper is connected to the output of the summing junction, and the input of the at least one digital IF processor is connected to the output of the error-shaped clipper.
Still further alternatively, the at least digital IF processor includes a plurality of digital IF processors, each of the plurality of digital IF processors having an input connected to the output of a respective CDMA modem, each of the plurality of inputs of the summing junction is connected to the output of a respective one of the plurality of digital IF processors, and the input of the at least one error-shaped clipper is connected to the output of the summing junction.
Other aspects and features of the present invention will become apparent to those ordinarily skilled in the art upon review of the following description of specific embodiments of the invention in conjunction with the accompanying figures.